Little is known about the epidemiology and prevention of medication errors and adverse drug events (ADEs) in the pediatric ambulatory setting. Children pose unique challenges because of weight-based dosing, the need for small nonstandard dosages, limited reserves to buffer errors, and limited communication skills to warn caregivers about symptoms of side effects or mistakes in medication administration. The major aims of the proposed study are to: 1) determine the rates, types, and predictors of medication errors and ADEs in the pediatric ambulatory setting, and 2) perform a randomized controlled trial (RCT) to assess the effectiveness of an intervention on reducing serious medication errors. The epidemiology of medication errors and ADEs will be determined in a two-month prospective cohort study of all patients under age 21 treated at six practice sites. Two sites in each of three types of communities will be studied: poor urban, working class suburban, and affluent suburban. Each practice site will be surveyed to capture important potential predictors of errors, such as level of training of healthcare providers, staffing, workload, and type of prescription refill system. Error detection and data collection will be performed using methods developed by our group, with appropriate adaptation for ambulatory pediatrics. All steps in the medication system will be monitored by reviewing duplicate prescriptions, conducting telephone surveys of families 10 days and three months following their clinic visit, and reviewing clinic charts four months after the visit. Medication errors and ADEs will be detected at all steps of this data collection process. Important demographic data that may predict errors will be collected, including racial, socioeconomic, educational, linguistic, and cultural barriers to understanding and complying with medication regimens. Two physician reviewers will independently determine if an ADE or potential ADE occurred and classify it on the basis of severity, preventability, and attribution. They will further determine which interventions could have prevented each serious medication error. This analysis of potential prevention strategies, and an additional analysis of predictors of serious medication errors based on data from aim 1, will inform the selection of an intervention to be tested in a RCT. The intervention will be introduced in a random manner in three clinics, one in each type of community, with the other three clinics serving as controls. Using the methods outlined above, we will collect data for six months at each clinic in order to determine the effectiveness of the intervention. We will additionally collect cost data to determine the cost-effectiveness of the intervention.